Association between obstructive sleep apnea and persistent postural-perceptual dizziness

1 Introduction

Persistent Postural-Perceptual Dizziness (PPPD) is a common cause of chronic dizziness. It is characterized by constant dizziness or unsteadiness lasting more than 3 months, with symptoms exacerbated by postural changes, head movements, complex visual stimulation or busy visual environments [21]. It is caused by long-term maladaptation to an initial dizzying trigger, and is often accompanied by anxiety, depression, and avoidance behaviours, which can lead to severe disability [21].

Patients with PPPD often have other somatic symptoms, including fatigue and overall increased sleepiness [24]. Previous work has shown a link between PPPD and sleep. Bittar et al. found that in a population of 81 patients diagnosed with PPPD, 38% of patients had sleep deprivation as a trigger for their PPPD [4]. More recently, Kim et al. showed that a majority of patients with PPPD screened positive for impaired sleep quality and for insomnia [17]. They further demonstrated that, among patients with different vestibular diagnoses, scores on a sleep quality index and on an insomnia index were highest in the PPPD group.

There have been a number of studies showing a possible association between obstructive sleep apnea (OSA) and vestibular dysfunction. Sowerby et al. showed that patients with idiopathic dizziness were more likely to have significant daytime somnolence than a group of controls with hearing loss but no dizziness [23]. Han et al. found abnormal vestibular caloric testing (unilateral or bilateral vestibular hyporeflexia) in 67.5% of a cohort of 77 OSA patients [14]. In that same study, bilateral vestibular dysfunction was associated with a lower lowest oxygen saturation (LSaO₂) on polysomnography. Kayabasi et al. found that patients with moderate-to-severe OSA had significantly higher rates of nystagmus and semicircular canal paresis than patients with mild OSA [16]. Similarly, OSA patients are significantly more likely to have abnormal vestibular caloric testing than healthy controls [13]. Patients with moderate-severe OSA also scored significantly higher than controls on the Dizziness Handicap Inventory (DHI), particularly on the physical subgroup [16]. Moreover, there is evidence that appropriate OSA treatment with continuous positive airway pressure (CPAP) can improve dizziness in patients with both OSA and Ménière’s disease [19].

Treatment of PPPD usually comprises a multi-faceted approach that may include changes in lifestyle, medications, vestibular rehabilitation, and psychotherapy. OSA is treatable, and if associated with PPPD, could represent a modifiable risk factor for PPPD. Despite much work on the relationship between sleep and vestibular dysfunction, there has yet to be a study investigating the possible link between PPPD and OSA.

This study was an exploratory look to determine whether there is an association between OSA and PPPD in a small group of patients. Another goal was to compare the utility of two validated OSA screening questionnaires (STOP-BANG and Berlin) in this patient population as each questionnaire was developed with a different population focus (perioperative setting and primary care, respectively). For those patients with OSA, we also sought to determine whether the degree of dizziness disability (as assessed by the Dizziness Handicap Inventory [DHI]) correlated with the severity of sleep apnea (as assessed by the Apnea-Hypopnea Index [AHI]).

2 Methods

3 Results

During the study period, 76 patients with a diagnosis of PPPD were approached for participation. Of these, 12 patients already had a diagnosis of OSA and were excluded from participation. Of the 64 patients meeting inclusion, 43 patients initially provided consent for participation. The most common reasons for declining participation at initial approach were logistical (distance from the tertiary center or residence in a different province). Twenty-five patients with PPPD completed the study and underwent PSG. The most common reason for loss to follow-up or withdrawal before PSG was the long wait time to PSG. Patients lost to follow-up or who declined PSG were not included in the final analysis.

The mean age of participants was 49±12.2, and they were 60% female (15/25) with mean BMI of 29.5± 6.3 kg/m². On average, the number of months between their initial dizziness ‘trigger’ for PPPD and presentation in clinic was 50 months. In terms of the actual triggers, in 8 patients vestibular migraine was felt most likely, followed by vestibular neuritis (5), BPPV (4), acute stress/anxiety (4), Ménière’s (1) and other (a central lesion, for example). Multiple triggers were recorded if felt equally contributory based on review of the medical record. At the time of assessment, 15 patients (60%) had comorbid migraine or vestibular migraine. Similarly, 17 patients (68%) were felt clinically to have stress or anxiety contributing to their current dizziness, though formal psychiatric assessment was not done. Except for one patient, all participants had normal bedside head impulse testing at the time of assessment. Of 14 patients who ultimately underwent formal vestibular testing, only 2 patients had abnormal vestibular function (both had unilateral caloric weakness).

The prevalence of OSA was 56% (14/25). The average AHI in those diagnosed with OSA was 18.5±10.2 events/hour (median, 17 events/hour). Patients with OSA were older (56±8 years versus 40±12.2 years) and had higher BMI (32±6 versus 26±5 kg/m²) and larger neck circumferences (15.5±1.7 vs 13.9±1.2 in) when compared to their non-OSA counterparts, but there was no clear gender bias [57% (8/14) versus 64% (7/11) female] (Table 1). When a cut-point of age 45 is used, only 11% of patients under 45 years had OSA. Conversely, 81% of patients age 45 or older in our study had OSA. Similarly, only 20% of patients with BMI less than 27 had OSA, compared with 80% of patients with BMI of 27 or greater. Of the five patients in the study who were both under 45 years and had BMI less than 27 kg/m², none had OSA.

DHI scores were similar between groups (68.6±13.6 versus 65.5±18.7, p = 0.63). Groups also did not differ in their Epworth Sleepiness Scores (5.9±6.2 versus 8.2±6, p = 0.22). Other PSG abnormalities in the whole group were infrequent: clinically significant periodic limb movements (2), poor sleep efficiency (4) or abnormal sleep architecture (8). Among those without OSA, 1 had periodic limb movements, 2 had poor sleep efficiency, and 3 had abnormal sleep architecture.

DHI and AHI showed a very weak negative correlation (R = –0.037, p = 0.9002). There was a weak negative correlation between DHI and nocturnal nadir O₂ saturation (R = –0.34, p = 0.2336). Neither correlation met statistical significance.

The mean sensitivity and specificity of the STOP-BANG questionnaire (with standard cut-off > =3) for detecting OSA was 86% and 55%, respectively. The mean sensitivity and specificity of the Berlin Questionnaire was 79% and 45%, respectively.

This is the first study, to our knowledge, to examine the prevalence of obstructive sleep apnea in patients with PPPD. OSA is highly prevalent in our small group, especially when compared with the best estimates of overall population prevalence (22% in men and 17% in women) [12]. This is true even when considering the higher mean BMI (32 kg/m²) in the OSA group (the prevalence of sleep apnea in females with BMI >30 has been estimated at 37.9% [22]). Patients in the OSA group were also older (56 vs. 35 years, p = 0.0006). Aside from age and BMI, patients with OSA in this study did not clearly demonstrate other typical OSA risk factors —for instance, they were mostly female and average neck circumference was 15.5 inches. For comparison, a point for neck size is given on the STOP-Bang questionnaire only for neck circumference greater than 16 inches for females and greater than 17 inches for males.

Both the STOP-BANG and Berlin questionnaires demonstrated good sensitivity (86% and 79%, respectively) as screening tools in this small group of patients. High-quality prospective studies have demonstrated clear benefit of treatment for patients with OSA, across a range of outcomes, from sleepiness to cognitive dysfunction to quality of life [6, 18]. These outcomes are all lifestyle factors that are critical in a treatment approach to PPPD. Since OSA is common and treatable, and polysomnography is non-invasive, our study hints at a role for screening all PPPD patients for sleep apnea, though more in-depth study is required. Of note, scores on the Epworth Sleepiness Scale (collected routinely prior to PSG at our sleep center) were not significantly different between groups, suggesting that screening based on symptoms of sleepiness alone may be insufficient. Clinical gestalt (screening for typical male OSA patients with large neck circumference) may also fail to identify PPPD patients with OSA, as they were predominantly female with smaller neck circumferences. In healthcare settings like ours, where wait times for PSG are long, a bedside screening tool or algorithm may help to better stratify who should go on to PSG. In settings where PSG can be obtained within weeks, we acknowledge bedside screening may be less useful if OSA is found to be very prevalent in PPPD.

Lastly, if OSA is ultimately confirmed as a risk factor for PPPD, there may be a role for screening all patients who experience an initial dizzying trigger (and are therefore at risk for developing PPPD). This would include patients with a range of vestibular disorders including vestibular neuritis, benign paroxysmal positional vertigo, vestibular migraine, and Ménière’s Disease. The overall goal would be to prevent progression from acute, treatable conditions to vestibular maladaptation and PPPD, which is constant and debilitating. Overall, our work highlights the need for larger, high-quality studies on the relationship between OSA and PPPD.

The direction of the association between OSA and PPPD remains an open question. Certainly, much recent work has shown that obstructive sleep apnea can have a direct impact on the vestibular system [14, 16]. This would support nocturnal hypoxia as a predisposing factor in the development of PPPD. A related disorder of vestibular maladaptation, Mal de Débarquement Syndrome, has been linked to poor sleep or sleep deprivation at onset of the disorder [7]. Restorative sleep may be crucial in enabling the vestibular system to adapt to change [3]. Alternatively, PPPD and OSA patients might simply share a common set of risk factors —such as high BMI. It could be the case that PPPD patients are less active on account of their symptoms and their avoidance behaviours, which predisposes them to higher BMI and thus to OSA. More work is needed to better elucidate the precise nature of the relationship.

We also note the high proportion of patients in this study with migraine or vestibular migraine (60%). Migraine and PPPD are highly co-morbid [4, 25], and vestibular migraine is often a key player in driving PPPD (both as a trigger and/or as a precipitating factor) [10]. Sleep quality in patients with migraine is generally poorer than in the general population [5], and sleep deprivation is a known trigger for migraine. One hypothesis is that poorly restorative sleep from OSA may exacerbate migraine or vestibular migraine, and thus put these individuals at a higher risk of developing chronic dizziness or PPPD.

In this small study, we were unable to demonstrate an association between more severe sleep apnea and worse dizziness-related quality of life. The Dizziness Handicap Inventory was very weakly correlated with Apnea-Hypopnea Index (R = –0.037) and weakly correlated with nadir O₂ saturation during the night (R = –0.34), but neither test met statistical significance. This finding may reflect any of the following possibilities: a lack of statistical power to demonstrate an association, true lack of relation (OSA can catalyze the development of PPPD but not in a dose-dependent manner), or ceiling effect (when PPPD symptoms appear, they only do so past a certain threshold severity).

The main limitations of this pilot study were its small size and large dropout rate (41.8%). The main reason for drop-out was the long wait for outpatient PSG. Many patients had lost interest in the study by time they were recalled for PSG many months later. We acknowledge that this is a source of potential selection bias, in that patients with subjective sleep complaints may have been more likely to remain in the study. The lack of a control group is also a limitation. Further work is needed to determine the prevalence of OSA in a large population of PPPD patients, to verify the usefulness of screening questionnaires in this population, as well as to determine the effect of treatment of OSA on symptoms of PPPD.

5 Conclusion

In this small, preliminary study on PPPD patients, the prevalence of OSA was much higher than in the general population. PPPD patients with OSA had higher BMI and were older, but they were mostly female and lacked other classic OSA risk factors like large neck circumference. Screening questionnaires appear to demonstrate good sensitivity to detect PPPD patients who would benefit from formal polysomnography. Overall, these findings hint at possible benefit to screening all PPPD patients for OSA. Future studies should confirm these findings and determine whether treatment of OSA improves symptoms in PPPD.